Content uploaded by Martin Ruef
Author content
All content in this area was uploaded by Martin Ruef on Jan 11, 2016
Content may be subject to copyright.
Strong ties, weak ties and islands:
structural and cultural predictors of
organizational innovation
Martin Ruef
How does the tendency of entrepreneurs to engage in innovation relate to their
structural and cultural embeddedness? Using micro-data on entrepreneurial teams
and the organizational innovations they attempt to develop, this article presents a
predictive model of creative action to address this question. Capacity for creative
action is seen to be a function of the ability of entrepreneurs to (i) obtain
non-redundant information from their social networks; (ii) avoid pressures for
conformity; and (iii) sustain trust in developing novel—and potentially profitable—
innovations. Probit analyses of over 700 organizational startups suggest that these
mechanisms exercise effects on innovation via the network ties and enculturation
of entrepreneurs.
1. Introduction
An understanding of organizational innovation is critical to analysts who seek to
account for industrial change. While considerable scholarly attention has been brought
to bear on the issue of innovation, much focuses on the diffusion of existing departures
from conventional routines or ideas rather than the creation of new routines or ideas
(for reviews see Rogers, 1995; Strang and Soule, 1998). Thus, social scientists have
considered how innovations are spread via direct communication (Coleman et al.,
1966), role equivalence (Burt, 1987), mesolevel networks (Hedström et al., 2000), and a
variety of institutional pressures (DiMaggio and Powell, 1983; Strang and Meyer, 1993).
These analyses have led to sophisticated models of the structural, spatial and temporal
dynamics of innovation diffusion (e.g. Strang and Tuma, 1993), but have not advanced
empirical understanding of how departures from established ideas and routines arise in
the first place.
Since Schumpeter (1934 [1911]), economic and organizational researchers have
also considered the introduction of new routines and ideas within the context of
industrial organizations (see review by Hage, 1999). Like the diffusion literature,
much of the research focus has not been placed on the initial appearance of innova-
tions per se but, rather, on the adoption of innovations or the patent protection and
introduction of viable innovations to a market. Empirical attention has emphasized a
variety of contextual variables, including firm age (Sørensen and Stuart, 2000), firm size
Industrial and Corporate Change, Volume 11, Number 3, pp. 427–449
© ICC Association 2002
(Damanpour, 1992), inter-organizational networks (Ahuja, 2000; Powell et al., 1996),
patent precedents (Podolny and Stuart, 1995) and technological regimes (Malerba and
Orsenigo, 1996).
The existing evidence on the diffusion of innovation and organizational adoption
of innovation leaves two key questions unanswered. First, how does the micro-level
context—both structural and cultural—of entrepreneurs contribute to their tendency
to deviate from established ways of thinking or doing things and lead to the
introduction of new innovations? In contrast to much prior research on diffusion, this
focuses attention on the initial appearance of innovations, seeking to avoid the
retrospective bias of studying only successful ideas and practices.
1
Second, how are
innovations tied to the emergence of novel organizations and organizational forms?
More specifically, what leads individuals to establish organizations that employ
radically new routines as opposed to organizations that simply reproduce established
ways of doing things (Aldrich, 1999: 80–81)? This emphasis breaks with the majority of
organizational research, which examines innovation within established structures, to
consider how those structures themselves evolve with entrepreneurial innovation.
In this article, these questions are explored using a framework that adapts existing
ideas on the structural embeddedness of economic action (Granovetter, 1985; see also
Uzzi, 1996, 1999). Generally speaking, the propensity among entrepreneurs toward
innovation as opposed to the reproduction of existing ideas is seen to be a function of
the types of social relationships those entrepreneurs are embedded within. Given that
much of the existing evidence on network tie strength (Granovetter, 1973; Bian, 1997)
and network diversity (Marsden, 1987: 124) focuses on dynamics of information
diffusion and social influence, a structural analysis predicting creative action requires
that the conventional view of embeddeness be amended. The theoretical framework
introduced below pays greater attention to role of network ties in inducing conformity
and sustaining trust, as well as spreading novel ideas.
The framework informs testable hypotheses that will be examined using an original
data set of 766 entrepreneurial teams attempting to start new business organizations.
Data were gathered on each of the teams regarding sources of ideas for their ventures,
previous career experiences, team structure, patent/trademark applications, and the
relative novelty of contributions they hoped to make with respect to a variety of inno-
vation categories. Statistical analyses employ probit models to explore how structural
features of the entrepreneur’s networks—strength of ties, diversity of ties and content
of ties—interact with features of the entrepreneurs’ career histories—work experiences,
enculturation and role diversity—to generate varying capacities for creative action.
Such creative action among the entrepreneurs manifests itself in a number of ways,
including the introduction of new types of products or services; the development of
new methods of production, distribution, or marketing; the development of new forms
1
Ibarra’s (1993) study of network characteristics and innovation roles is an important exception to the
general lack of micro-level analyses of innovation. However, it predicts an individual’s role in an
innovation process rather than the occurrence of innovation per se.
428 M. Ruef
of external linkages; entry into unexploited market niches; and the restructuring of
industries.
2. Theoretical framework
Table 1 summarizes the basic model of organizational innovation. I link creative action
to three underlying mechanisms, including the ability of entrepreneurs to access diverse
sources of information, to avoid pressures for conformity and to sustain trust with
others who are told about a potential innovation. Within this framework, the
propensity of individual entrepreneurs to break with convention is both encouraged by
social relations—which may bring disparate ideas, routines or technologies to an
entrepreneur’s attention—and discouraged by social relations—which may introduce
pressures for conformity or concerns about trust. To use Giddens’s (1984) terminology,
social structure is seen as both enabling and constraining. The balance of tensions
toward and away from innovation is largely determined by aspects of an individual’s
relational context: the strength, diversity and content of network ties. These tensions
are not simply seen as reflections of the present structural embeddedness of
entrepreneurs, but also of their cultural embeddedness, reflecting a past history of work
and educational relationships.
2.1 The strength of weak ties revisited
Granovetter (1973: 1361) defined the concept of a strong interpersonal tie in terms of
the time and emotions invested in a relationship, as well as the reciprocity involved
between participating actors (see also Marsden and Campbell, 1984). Typical examples
of strong ties include friendship and familial relationships. Weak ties, by contrast, entail
Table 1 Predictors and underlying mechanisms of organizational innovation
Information Conformity Trust Innovation Relevant
hypotheses
Structural predictors
Strong ties 0 + + 0
Weak ties + 0 0 + H1, H7–H8
Directed ties (mimetic) + + n/a 0 H2
Directed ties (expert discourse) + 0 n/a + H3
Network diversity + + + + H4
Team size + 0 0 + H5
Cultural predictors:
Team diversity + – (–) + H6
Industry LOT 0 + n/a – H9
Strong ties, weak ties and islands 429
more limited investments of time and intimacy, subsuming an array of social
acquaintances. Granovetter’s (1973, 1995) influential ‘strength-of-weak-ties’ thesis
maintains that weak ties are often more important in spreading information or
resources because they tend to serve as bridges between otherwise disconnected social
groups; strong ties lead to less efficient transmission processes because a large number
of actors in the strong tie network also know each other, as well as knowing the focal
actor. While other network analysts have pointed out that strong ties can also serve as
bridges (Burt, 1992: 27–30), their tendency to be redundant sources of information or
resources remains a widely accepted tenet of structural theory.
To clarify the relevance of tie strength for innovation, rather than simple diffusion,
it will be useful to consider two underlying dimensions of social relationships:
information and influence. Schumpeter (1934 [1911]) describes innovative action as
the novel combination of existing ideas and routines. Innovation thus requires, first
and foremost, that corporate actors have access to information on disparate ideas and
routines from which elements can be combined. Consistent with the ‘strength-
of-weak-ties’ thesis, both the transmission rate and availability of such disparate
information will be higher for individuals relying on weak network ties rather than
strong ties.
The impact of influence on innovative propensity is more complex. Recent research
on job searches has suggested that influence—entailing pressures for conformity—may
explain why strong ties can be more important for status attainment purposes than
weak ties (Bian, 1997). The dynamics of influence are quite different, however, when the
outcome of interest involves innovative action. The most pertinent influence affecting
innovation is not that directed from a focal actor to a set of others, but, rather, that
directed from other actors in a network to the focal actor. Although the former
influence is relevant to the diffusion and (perceived) success of an innovation, the latter
is central to the initial departure from established conventions. Strong ties impose
greater demands for conformity on a focal actor. Family members and friends who are
consulted regarding new business ideas may be insulted when other elements are
introduced that deviate from or clash with their own way of doing things. The affective
content of these relationships strengthens the role of their influence, since actors are
expected to heed the advice of family and friends. Weak tie relationships, on the other
hand, allow for more experimentation in selectively combining ideas from one source
with those of another and impose fewer concerns regarding social conformity. The
combined effect of information diffusion and influence on innovative propensities thus
yields the following hypothesis:
Hypothesis 1. Actors relying on strong ties as sources of ideas are less likely
to be innovative than actors relying on weak ties.
2.2 Islands: the strength of directed ties
The problem of conformity that is inherent in many social relationships raises the
question of whether relatively isolated entrepreneurs might not be capable of even
430 M. Ruef
greater innovation than those embedded in weak tie networks. In other words, does the
atomistic model of Homo economicus or Homo faber, the lone tinkerer, ring true for the
most radical of departures from extant routines? While such insular depictions do seem
to free actors from pressures for conformity, they fail to acknowledge the importance of
information flows as inputs to innovation. As Uzzi’s (1996, 1999) recent studies of
business networks have pointed out, the paradox of embeddedness is that economic
action often benefits from initial increases in relational ties, but suffers when actors are
highly embedded. Similarly, discussions of innovation within established firms suggest
the importance of loose coupling among business units, rather than tight coupling or
decoupling (Saloner et al., 2001: chap. 5).
For individual innovators, one possible solution to the structural dilemma involves
directed ties. Such ties attempt to separate the need for information from the constraints
of influence. Instead of entailing a social relationship—where two actors are reci-
procally directed toward each other—directed ties involve a unilateral monitoring of
discourse and activities on the part of other actors (see Weber, 1968 [1924]). Network
analyses (e.g. Boorman, 1974) suggest that such social structures—so-called ‘islands’,
with limited communication among subpopulations—can prove important in
sustaining innovations that deviate from majority views. In Boorman’s mathematical
model, the ability of a novel social trait to survive (and, ultimately, take over) depends
on its selective isolation from a broader population of social traits. Similarly, isolated
entrepreneurs with directed ties to a variety of information sources can be charac-
terized in terms of ‘island’ network structures, which may encourage the combination
of disparate ideas without necessarily imposing demands for conformity.
Whether directed ties actually do lead to innovation depends on the specific
character of unilateral monitoring. Scholars examining processes of imitation among
entrepreneurs and organizations have long recognized the propensity of some directed
ties to produce isomorphism rather than innovation (DiMaggio and Powell, 1983;
Miner et al., 1999). An entrepreneur whose attention is directed at relatively concrete
implementations of ideas and routines (often bundled in the form of extant formal
organizations) may reproduce those competencies with minimal modification. At the
same time, when ideas and routines are abstracted from specific industries—or
‘theorized’, as institutional scholars have put it (Strang and Meyer, 1993)—then the
potential for creative modification of those ideas is enhanced considerably. Insular
individuals who direct their attention to public discourse in a topic area are especially
well-suited to be innovators for a number of reasons: (i) new information often diffuses
more rapidly via such media sources than through interpersonal relationships (Strang
and Meyer, 1993); (ii) the unilateral nature of the relationship (‘directed’ to an expert or
a text) does not impose the same demands for conformity as a reciprocal relationship;
and (iii) the generalized content of the discourse means that considerable experi-
mentation is often necessary to adapt ideas to the particular circumstances of the
entrepreneur or their organization. While the first two mechanisms primarily ensure
rapid, unencumbered diffusion of ideas, the third seems more fundamental to
Strong ties, weak ties and islands 431
innovation: as entrepreneurs wrestle with abstract ideas presented in the business press,
in technical papers or in educational coursework, they are forced to modify and
‘ground’ them in the specific problems and prospects of their own industries. In sum-
mary:
Hypothesis 2. Actors relying on ties directed to the concrete activities of
other individual or corporate actors are less likely to be innovative than
actors relying on weak ties.
2
Hypothesis 3. Actors relying on ties directed to the abstract discussion of
ideas in expert discourse are more likely to be innovative than actors relying
on weak ties.
2.3 Network diversity
Entrepreneurs depending exclusively on one type of network tie or another—strong,
weak or directed—can be rank-ordered readily in terms of their propensity toward
innovation. However, many innovators derive their ideas from multiple sources, often
involving a mixture of family members, friends, acquaintances, or individuals and firms
with whom they have had no prior contact. Studies of business startups among nascent
entrepreneurs (Renzulli et al., 2000) suggest that such network diversity can have
beneficial effects above and beyond the cumulative effect of networks ties considered
individually. By the same token, diversity can be seen as an impetus to creative action
insofar as the relative advantages of each type of tie (with respect to information and
influence) tend to offset the disadvantages of other ties.
Ties directed to the discourse of experts generally have the most favorable structural
properties for information diffusion, considering the lack of relational intermediaries
and consequent fidelity of the information received. At the same time, directed ties do
not include any mechanism for iterative feedback with respect to a potential
innovation. This means that actors relying exclusively on such ties will tend to adapt
ideas to their own circumstances and needs without necessarily engaging in further
innovation. Those actors that are also embedded in weak or strong tie networks, on the
other hand, may subject new combinations of ideas to further modification, as they
receive feedback on the innovation from family, friends and acquaintances. This is
especially true for strong tie networks, in which entrepreneurs are able to place trust in
other actors and have some confidence that novel ideas will not be coopted.
The drawbacks of conformity in strong and weak tie networks can also be
ameliorated by diversity. The influences of family members or friendship circles are
strongest when a focal actor has relatively limited access to other sources of ideas.
Pressures toward conformity are offset to some extent when an entrepreneur is also
able to gather insights from more casual acquaintances and claims made in public
discourse. More generally, as Burt (1992: 195) points out, individuals who enjoy greater
2
In particular, this is likely to be true when the actors being observed are active within ego’s own
industry. Adoption of ideas from other industries typically entails greater experimentation.
432 M. Ruef
heterogeneity in their role relationships can ‘be less concerned about getting [their
own] role “right”’.
Hypothesis 4. Actors embedded in a diverse set of network ties are more
likely to be innovative than actors relying on homogenous ties.
2.4 Team structure and internal ties
The discussion of structural embeddedness thus far has continued to maintain one bias
of undersocialized conceptions of action—that of the solo entrepreneur. While the
concept of the solo entrepreneur is convenient from the standpoint of classical
economic models, it has been superseded by the recognition that innovation is often
undertaken by entrepreneurial teams (e.g. Stewart, 1989). Aside from considering the
external ties within which actors are situated, the analysis of creative action must also
address aspects of team structure and the nature of internal network ties.
In this respect, perhaps the simplest predictor of innovative propensity is team size.
New combinations of ideas are encouraged when a number of entrepreneurs choose to
work together and apply multiple perspectives to a problem. Solo entrepreneurs, by
contrast, are more likely to reproduce familiar routines from their own life experiences.
Beyond team size, the diversity of functional roles represented among one or more
entrepreneurs is likely to have a marked impact on the creativity of action. A team
composed of members with accounting, marketing and engineering backgrounds is
more likely to produce new combinations of ideas than a team composed only of
accountants. Similarly, a solo entrepreneur with backgrounds in both marketing and
engineering may combine ideas from these disciplines in novel ways. Burt (1992:
18–20) suggests that the lack of role equivalence among entrepreneurs (or their
contacts) is often as important in obtaining non-redundant information as avoidance
of cohesive ties. The multiplicity of role structures that individuals or entrepreneurial
teams may invoke can therefore contribute substantially to deviations from established
ways of doing things.
Hypothesis 5. Large entrepreneurial teams are more likely to be innovative
than small teams or solo entrepreneurs.
Hypothesis 6. Entrepreneurial teams or actors drawing on a diverse set of
functional roles are more likely to be innovative than those drawing on
homogenous roles.
The information-theoretic benefits of entrepreneurial teams need to be balanced
with the recognition that such groups, like external network ties, can also impose some
demands for conformity on their members. When the internal ties among team
members reflect limited prior acquaintance, this influence tends to be small and the
benefits of team structures are maximized. Innovative propensity is likely to be reduced,
however, when there have been some prior ‘arm’s length’ relationships between team
members (‘weak ties’) and reduced even more when team members know each other
Strong ties, weak ties and islands 433
well (‘strong ties’). Naturally, the benefits of low structural embeddedness only accrue if
we assume that the problem of trust is overcome within entrepreneurial teams by
providing each team member with an equity stake in the venture.
Hypothesis 7. Entrepreneurial teams composed of weak ties are less likely
to be innovative than teams involving members with limited prior
acquaintance.
Hypothesis 8. Entrepreneurial teams composed of strong ties are less likely
to be innovative than teams involving members with weak tie relationships.
2.5 Cultural embeddedness
Cultural embeddedness reflects the amount of experience that actors have in a
particular task domain, the extent to which they consciously draw ideas from that
experience, and whether the experience involved conventional routines and com-
petencies or attempts to deviate from conventions. Actors (or entrepreneurial teams)
with extensive experience in an industry are less likely to be innovative than those with
limited experience. Examining organizational innovation, researchers have noted that it
is ‘indifference to industry routines and norms [that] gives an outsider the freedom to
break free of the cognitive constraints on incumbents’ (Aldrich and Kenworthy, 1999:
20). Moreover, the performance of experienced participants in an industry tends to
become increasingly predictable and reliable. These features of performance are
generally valued by society (Hannan and Freeman, 1984), but they can also inhibit
entrepreneurial exploration (March, 1991). As Sewell (1992: 18) emphasizes, the
unpredictability of performance—especially, on the part of inexperienced actors—is a
key predictor of creative action, since it contributes to the reconsideration of estab-
lished cultural schema. Consequently:
Hypothesis 9. Actors (or entrepreneurial teams) with extensive experience
in an industry are less likely to be innovative than those with limited
experience.
When this proposition is applied to multi-member teams, there is one clear com-
plication, insofar as the dispersion of industry experience on the part of members—as
well as average—may influence innovation rates. If cohort effects are crucial, then the
level of innovation will depend on whether all team members entered the industry
around the same time period or during different periods. This issue is explored
empirically in the analyses below.
3. Data, measures and methodology
3.1 Data
Schumpeter (1934: 66) noted that innovation is often embodied in the creation of new
formal organizations, a viewpoint that has been echoed in studies of organizational
434 M. Ruef
demography (Carroll and Hannan, 2000). The present analysis pursues this insight by
examining innovation among a sample of 766 entrepreneurial teams who attempted to
start business ventures. The entrepreneurs were drawn from a population of business
professionals receiving MBA (masters of business administration) degrees from a
graduate program in the western United States. The sampling frame explicitly controls
for the wide variety of educational experiences and business competencies typically
found among nascent entrepreneurs (see Reynolds and White, 1997), but also limits the
representativeness of the entrepreneurs studied. Given that the emphasis of the present
study is on how structural and cultural embeddedness affect innovation—rather than
on the impact of individual traits—this trade-off appears to be justified.
In 1999, surveys were sent to the 1786 individuals in the sampling frame who were
identified as entrepreneurs in alumni records—with ‘entrepreneurs’ being defined as
individuals who had tried to start a business, either as a primary or secondary career
activity. I received 766 non-duplicate surveys, yielding a response rate of nearly 43%.
3
The respondents were asked to identify the nature of their entrepreneurial effort, the
types of innovations they hoped to introduce, primary sources of business ideas, the
composition of their founding team, and any external partners or advisors that may
have been involved in the founding process. Where possible, secondary data sources
were used to confirm details of startup activities. Additional tests of measure reliability
were applied, as discussed below.
Because the data were generally collected for a given period in time (organizational
startup), some precautions were taken to minimize problems of endogeneity. To ensure
causal precedence, the factors contributing to the structural and cultural embeddedness
of entrepreneurs should pre-date attempts at innovation. Such causal precedence was
established exhaustively for the industry experiences of the entrepreneurs, based on
surveys of their career histories. Moreover, 88% of valid respondents reported formal
network-building activities (e.g. founding team formation) prior to, or simultaneously
with, formal innovation activities (e.g. patent or trademark applications). While
endogeneity cannot be ruled out in the remaining cases, it does appear that the
structure of teams and entrepreneurial networks pre-dated innovative propensity in
many of these startup efforts.
4
3.2 Dependent measure
Given the inherent difficulties in measuring innovative action, I employed two
operationalizations of the dependent variable—one emphasizing objective behaviors
on the part of entrepreneurs and the other emphasizing subjective perceptions of
innovation. Following previous studies of innovation, the behavioral measure examines
patent and trademark applications advanced by an entrepreneur during the creation of
3
Three duplicate surveys were used to validate information, but were subsequently removed from the
sample to ensure that each team contributed only one case to the analysis.
4
Clearly, repeated measures of innovation—coupled with data on changes in network structure—are
desirable in addressing the endogeneity issue more directly.
Strong ties, weak ties and islands 435
a new startup.
5
The decision to sample applications, in particular, rather than patents or
trademarks issued, hinges on the potential success bias among startups that manage to
legally protect their ideas or routines. Patent protection may be as much a function of
the resources and stakeholders that are backing a startup, as the actual creativity of ideas
involved.
6
At the same time, the input of legal and management counsel during the
patent/trademark application process renders this indicator less subjective than
individual perceptions of innovation.
Subjective perspectives on innovation consider the opinions of participants in an
institutional arena, including experts and entrepreneurs. While the opinions of experts
seem a likely candidate for judging innovation, they suffer from one major short-
coming: the attention of many experts—e.g. industry specialists, stock market analysts,
academics—is directed largely at successful instances of creative action. There is a
considerable risk that their assessments of creativity may be conflated with assessments
of success. Moreover, expert evaluations are not publicly available for large numbers of
unsuccessful innovations.
An alternative measure of innovation focuses on the perspective of entrepreneurs
themselves—to what extent are they attempting to combine disparate ideas or routines,
independently of the success of those combinations? This is the second measurement
strategy employed in the present analysis, considering categories of economic
innovation that elaborate on Schumpeter’s widely used approach (see Schumpeter,
1934). Nine categories are addressed, including: (i) the attempted introduction of a new
type of product/service in a local or regional market niche; (ii) the attempted
introduction of a new type of product/service in a national or international market
niche; (iii) the attempted introduction of a new method of production, (iv) distri-
bution or (v) marketing; (vi) the development of new supplier linkages; (vii) attempted
entry into an unexploited market niche; (viii) reorganization of an industry or
organizational population; and (ix) a residual category of innovations identified by the
entrepreneurs. Any number of these categories can be applied to a given organizational
startup. Table 2 summarizes the frequencies with which the sampled entrepreneurs
aligned their ventures with various attempts at innovation.
7
5
As Damanpour (1988) notes, other a priori sampling designs that impose the investigators’ definitions
of innovation are likely to miss significant aspects of creative action.
6
For an embedded theory of innovation, one of the principal difficulties with the conflation of
innovation and success is that it becomes very difficult to disentangle the effects of information and
resource flows. Thus, research on successful organizational patent introductions (e.g. Ahuja, 2000) is
forced to jointly consider knowledge spillover and resource sharing among firms.
7
The nine categories of innovation are logically independent, with the exception of (i) and (ii).
Product/service entry into a national or international market (ii) subsumes entry into regional and
local markets (i). This operationalization allows considerations of geographic scope to weigh into the
index of innovation: e.g. introducing a new product that has not been seen before in the global market
is considered to be more innovative than introducing a product that has not been seen before in a
regional market.
436 M. Ruef
For each entrepreneurial team, an ordinal scale was derived by summing the number
of applicable innovation categories. This subjective scale—reflecting how innovative
the entrepreneurs believed their organizational startup to be—was then analyzed
further for construct validity. Rank orderings on the scale were compared across two
categories of sampled organizations known to have very different propensities for
innovation: franchises and ‘true’ startups. By definition, franchises tend to reproduce
existing organizational templates and routines, introducing few innovations in the
process. A comparison of rank sums for the two subsamples confirms that the sub-
jective scale clearly picks up on this distinction, with franchises being significantly less
Table 2 Descriptive statistics for entrepreneurial teams (total n = 766)
Variable Mean/proportion Valid n
Innovation index 2.16 737
Introduce new product/service type 0.56
—into national or international market 0.34
Introduce new production method 0.09
Introduce new distribution method 0.13
Introduce new marketing method 0.19
Develop new supplier linkages 0.11
Enter unexploited niche 0.51
Reorganize organizational population 0.10
Other Innovation 0.14
Patent or trademark application filed 0.30 737
Structural embeddedness
Strong ties 0.38 753
Weak ties 0.52 753
Directed ties (discourse) 0.19 753
Directed ties (mimetic) 0.33 753
Network diversity 0.16 753
Cultural embeddedness
Years of experience 9.88 766
Reliance on entrepreneurial experience 0.52 766
Reliance on work experience 0.39 753
Team composition
Number of entrepreneurs 2.28 766
Role diversity 2.89 766
Strong ties
a
0.52 437
Weak ties 0.30 437
a
Tie composition is only applicable for teams with more than one member; ‘no ties’ represents
the omitted structural category.
Strong ties, weak ties and islands 437
innovative than other startups at the P < 0.001 level (Mann–Whitney U = 1848.5;
one-tailed test).
Construct validity was also considered by analyzing the pairwise correlation between
the two operationalizations of the dependent variable. A significant, though moderate,
level of correlation was expected, insofar as the behavioral and subjective measures tap
into somewhat distinctive aspects of innovation—i.e. application for patent protection
primarily addresses product and method innovations within the subjective scale
(categories i–v, above) and trademarks are often associated with ‘branding’ efforts in
unexploited markets (category vii). Consistent with this expectation, the non-
parametric (Spearman) correlation of the behavioral indicators with the subjective
scale is 0.32 and statistically significant at the P < 0.001 level.
3.3 Independent measures
Table 2 lists the descriptive statistics for all of the independent variables employed in the
analyses, reflecting the embeddedness of the entrepreneurs, the composition of their
teams, and some control variables. Listwise deletion of missing values produced 730
cases for the analysis of all entrepreneurs and 421 cases for the analysis of multimember
entrepreneurial teams.
Structural embeddedness. Individual entrepreneurs and teams were asked to identify
what sources inspired their initial business idea, using a non-mutually-exclusive coding
scheme. The sources were classified into four categories of structural embeddedness:
(a) discussions with family members or friends (‘strong’ ties); (b) discussions with
business associates, such as customers or suppliers (‘weak’ ties); (c) discussion in the
general media or specialized trade press (ties ‘directed’ toward discourse); and (d)
observation of existing competitors in an industry (ties ‘directed’ toward a set of
concrete others). Descriptive statistics suggest substantial reliance on weak ties (52% of
the cases), consistent with the prevalence of such ties in status attainment processes
(Granovetter 1995). Reliance on strong ties was more limited (38%), while com-
paratively few respondents indicated that business ideas developed from attention to
the discourse of experts (19%).
Network diversity was calculated based on the teams’ list of external partners,
advisors and other sources of information. Prior relationships with these sources were
again classified in terms of strong ties, weak ties and no prior contact. For purposes of
calculating network diversity, family members were distinguished from friends within
the category of strong ties. Diversity (D) was then computed in terms of an information
entropy measure proposed by Shannon and Weaver (1963):
(1)
where n is the number of categories for social ties and y
i
is the proportion of others
D
y
n
y
i
i
n
i
=−
F
H
G
I
K
J
=
∑
log
log
1
438 M. Ruef
listed by the respondent within each category i. The measure varies from 0 for com-
pletely homogenous networks to 1 for completely heterogeneous networks.
Cultural embeddedness. The enculturation of entrepreneurs in a task domain is ex-
pected to increase with industry tenure. For purposes of our analysis, this experience is
measured by the average number of years that entrepreneurs have been in an industry.
8
Among entrepreneurial teams, the dispersion—as well as mean—of this variable is of
interest, since it may tap into the extent to which team members come from different
cohorts and are familiar with different product/service paradigms.
Team members were also asked whether ideas from previous work experience were
used in developing their new venture and whether the team included any serial
entrepreneurs. As shown in Table 2, 39% of the teams imported ideas from other work
contexts, while 52% could rely on the previous entrepreneurial experience of one or
more of their members.
Entrepreneurial teams. The size distribution of the teams reveals that 65% include more
than one member, with 2.28 members representing the mean for the sample. Role
diversity is measured as the number of business functions filled within the teams,
considering the following specialties: accounting; finance; human resources; legal;
marketing; operations; business strategy; information systems; engineering; and
research and development. On average, the sampled individuals and entrepreneurial
teams subsume a little under three of these functional roles.
For multimember teams, data were also collected on the prior relationships con-
necting team members. Those teams composed exclusively of family members, friends
or work colleagues were classified as consisting of a set of ‘strong’ ties. Those teams
composed exclusively of members with no prior contact with one another (often
introduced via an intermediary, such as a venture capitalist) were classified as having
no ties. Those teams composed of acquaintances—or a mixture of family members,
colleagues, acquaintances, etc.—were classified as consisting of a set of ‘weak’ ties.
Interestingly, the majority of multimember teams fall into the ‘strong’ ties category,
suggesting that the requirements for trust and emotional support within the team are
much more pronounced than they are with respect to external advisors or sources of
information.
Control variables. Fixed effects were included in all models to address the fact that
propensity toward innovation may vary by industry. Many economists have maintained
that organizational innovation rates in service sectors are lower (or lead to fewer
increases in productivity) than in manufacturing sectors (cf. Sundbo, 1998: 99–102). I
distinguish among eight industry categories, organized broadly by Standard Industrial
8
It could be argued that interpreting the effects of industry tenure in terms of enculturation requires
that the confounding effect of the entrepreneurs’ ages be controlled for. I conduct a sensitivity analysis
below to ensure that the impact of industry experience on innovation is distinct from the chronological
maturity (and possible risk aversion) of the entrepreneurs.
Strong ties, weak ties and islands 439
Classification (SIC) (see Table 3). Extractive industries (agriculture and mining) serve
as the reference category in the analyses.
3.4 Methodology
Given that the subjective scale of innovation is measured on an ordered scale,
multivariate methods for ordinal variables were applied to predict the level of perceived
innovation within each entrepreneurial team. Zavoina and McElvey’s (1975) ordered
probit model was estimated based on the following specification, stated in terms of a
continuous latent measure of innovation (Y*):
(2)
where Y is the observed counterpart to Y* and the µs are free threshold parameters that
distinguish different ordered values. Maximum likelihood estimates were derived using
Greene’s (1998) LIMDEP software, with the constraint that the first threshold para-
meter (µ
0
) equal zero.
Applications for patents or trademarks by entrepreneurial teams were treated as
dichotomous outcomes (whether or not an application was filed as part of the teams’
startup activities). A simple probit model was applied in these cases, again employing
maximum likelihood estimation.
YX Y Y
Y
Y
JY
J
**
*
*
*
=+ = ≤
<≤
<≤
>
−
βε µ
µµ
µµ
µ
with if
if
if
if
0
1
2
0
01
12
1
K
Table 3 Distribution of represented industries (valid n = 737)
Industry Standard industrial
classification
No. of firms % applying for
patent/trademark
Agriculture and mining divisions A and B 15 13
Construction division C 76 7
Manufacturing (electronic and
computer components/industrial
machinery)
division D, major groups
35 and 36
69 64
Manufacturing (other) division D, major groups
20–34, 37–39
64 58
Transportation and utilities division E 40 20
Wholesale/retail trade divisions F and G 107 37
Finance, insurance and real estate division H 132 10
Other services division I 234 30
440 M. Ruef
4. Results
The effect of structural and cultural embeddedness on creative action was analyzed in
three stages. Table 4 considers the entrepreneurs’ own perceptions of innovation,
reporting estimates for all sampled entrepreneurs (n = 730 after listwise deletion). Table
5 develops a parallel series of models for the behavioral indicator of innovation–
patent/trademark application. Table 6 summarizes selected parameter estimates
pertaining to the effect of team composition, considering multimember teams only (n
= 421 after listwise deletion). The reference category for all comparisons is a simple
caricature of Homo economicus—an isolated entrepreneur, with no social ties (directed
or undirected) to external sources of information, no prior enculturation in an
industry and no entrepreneurial team.
Table 4 Ordered probit models predicting subjective perception of innovation based on
experience, embeddedness and team structure of entrepreneurs
Variable Model 1 Model 2 Model 3 Model 4
Intercept
a
2.019 (0.269)*** 1.958 (0.278)*** 1.903 (0.283)*** 1.627 (0.279)***
Industry context
b
Construction –0.663 (0.280)* –0.604 (0.284)* –0.600 (0.286)* –0.541 (0.277)
Manufacturing (computers/electrical
components)
–0.031 (0.319) –0.007 (0.317) –0.056 (0.316) –0.098 (0.304)
Manufacturing (other) 0.116 (0.286) 0.168 (0.292) 0.124 (0.295) 0.154 (0.285)
Transportation and utilities –0.208 (0.292) –0.263 (0.298) –0.328 (0.306) –0.295 (0.304)
Wholesale/retail trade 0.211 (0.259) 0.192 (0.267) 0.164 (0.272) 0.228 (0.265)
Finance and insurance –0.563 (0.264)* –0.558 (0.268)* –0.516 (0.271) –0.444 (0.263)
Other services –0.274 (0.254) –0.289 (0.259) –0.282 (0.262) –0.200 (0.254)
Cultural embeddedness
Years of experience –0.012 (0.004)** –0.011 (0.004)** –0.010 (0.004)** –0.009 (0.004)**
Reliance on entrepreneur’s experience 0.040 (0.140) –0.008 (0.140) –0.057 (0.140) –0.191 (0.142)
Reliance on work experience –0.156 (0.144) –0.223 (0.145) –0.260 (0.145) –0.297 (0.144)*
Structural embeddedness
Strong ties — –0.080 (0.084) –0.101 (0.085) –0.091 (0.086)
Weak ties — 0.230 (0.085)** 0.229 (0.086)** 0.227 (0.087)**
Directed ties—discourse — 0.300 (0.101)** 0.242 (0.101)** 0.240 (0.103)**
Directed ties—mimetic — –0.061 (0.093) –0.075 (0.094) –0.099 (0.094)
Network diversity — — 0.993 (0.183)*** 0.903 (0.182)***
Team Composition
Number of entrepreneurs — — — 0.071 (0.037)*
Role diversity — — — 0.059 (0.031)*
–2 log likelihood 2359.86 2340.95 2303.64 2287.48
Number of cases 730
a
Inclusion of intercept requires that m
0
be constrained to zero.
b
Extractive industries (agriculture/mining) represent the omitted category.
*P < 0.05; **P < 0.01;***P < 0.001 (one-tailed tests for hypotheses; two-tailed tests otherwise).
Strong ties, weak ties and islands 441
4.1 Perception of innovation
Cultural embeddedness has a substantial impact on the perception of innovation
among entrepreneurs (see Table 4, Model 1). Every year of industry tenure decreases
the likelihood that entrepreneurs will introduce what they consider to be fresh
organizational ideas in their startups. Following Hypothesis 9, it can be suggested that
this reflects both the increasing predictability of means—given the internalization of
standardized routines and competencies—and the increasing predictability of per-
formance with industry tenure. As individuals become socialized in a task domain, they
are less likely to change organizational methods they view as either appropriate or
effective. Sensitivity analyses (not shown) indicate that this effect continues to hold
when the chronological ages of the entrepreneurs are entered into the model. Other
control variables related to cultural embeddedness—such as the tendency of entre-
preneurs to draw on previous work or entrepreneurial experiences—are not statistically
significant.
Table 5 Probit models predicting likelihood of patent/trademark application based on
experience, embeddedness and team structure of entrepreneurs
Variable Model 1 Model 2 Model 3 Model 4
Intercept –2.160 (0.825)** –2.262 (0.841)** –2.342 (0.846)** –3.003 (0.867)***
Industry context
a
Construction –0.587 (0.898) –0.529 (0.899) –0.510 (0.901) –0.408 (0.904)
Manufacturing (computers/electrical
components)
2.229 (0.811)** 2.271 (0.812)** 2.259 (0.814)** 2.182 (0.816)**
Manufacturing (other) 2.592 (0.815)** 2.627 (0.817)** 2.617 (0.819)** 2.705 (0.822)**
Transportation and utilities 0.429 (0.865) 0.332 (0.868) 0.301 (0.870) 0.304 (0.877)
Wholesale/retail trade 1.358 (0.796) 1.304 (0.796) 1.297 (0.798) 1.408 (0.801)
Finance and insurance –0.353 (0.827) –0.371 (0.827) –0.325 (0.829) –0.277 (0.835)
Other services 1.101 (0.783) 1.086 (0.782) 1.105 (0.784) 1.265 (0.788)
Cultural embeddedness
Years of experience –0.032 (0.011)** –0.030 (0.011)** –0.030 (0.011)** –0.029 (0.011)**
Reliance on entrepreneur’s experience 0.868 (0.334)** 0.819 (0.336)* 0.795 (0.337)* 0.505 (0.346)
Reliance on work experience 0.105 (0.348) 0.032 (0.351) 0.019 (0.352) –0.037 (0.356)
Structural embeddedness
Strong ties — –0.052 (0.192) –0.061 (0.193) –0.025 (0.196)
Weak ties — 0.173 (0.190) 0.174 (0.190) 0.183 (0.193)
Directed ties—discourse — 0.495 (0.224)* 0.460 (0.226)* 0.458 (0.231)*
Directed ties—mimetic — –0.079 (0.198) –0.089 (0.198) –0.130 (0.201)
Network diversity — — 0.562 (0.364) 0.336 (0.376)
Team composition
Number of entrepreneurs — — — 0.259 (0.078)***
Role diversity — — — 0.051 (0.067)
–2 log likelihood 736.89 730.98 728.61 710.24
Number of cases 730
a
Extractive industries (agriculture/mining) represent the omitted category.
*P < 0.05;**P < 0.01;***P < 0.001 (one-tailed tests for hypotheses; two-tailed tests otherwise).
442 M. Ruef
The addition of covariates assessing the structural embeddedness of entrepreneurs
improves model fit substantially (Model 2 vs. Model 1, likelihood ratio χ
2
= 18.91, ∆d.f.
=4,P < 0.001). Consistent with the ‘strength-of-weak-ties’ proposition (Hypothesis 1),
entrepreneurs who rely heavily on information from acquaintances are more likely to
engage in what they see as innovative activity than those who rely on information from
family and friends {odds ratio = exp[0.23 — (–0.08)] = 1.36}. The reduction of
information redundancy and conformity in weak-tie networks creates a milieu where
attempts at creative action are more likely than in strong-tie contexts. The impact of
conformity may be reduced further when entrepreneurs direct their attention to claims
made in public discourse, rather than to the ideas of acquaintances (Hypothesis 2).
Ceteris paribus, these entrepreneurs are nearly 1.5 times as likely to view themselves as
innovators than those who rely on strong ties and slightly more likely (1.1 times, not
significant) than those who rely on weak ties. By contrast, entrepreneurs who direct
their attention to the concrete behaviors of existing competitors in an industry, rather
than the discourse of experts, tend to reproduce conventional routines and ideas
(Hypothesis 3). While errors in the mimetic process can ultimately lead to unforeseen
innovation (Alchian, 1950), these mutations lie outside the initial subjective purview of
the entrepreneurs.
A third model explores the impact of network diversity on perceptions of creative
action. Entrepreneurs who are embedded in heterogenous networks, comprising a
mixture of strong ties, weak ties and advisors with no prior relationship, are
significantly more likely to attempt innovation than entrepreneurs in homogenous
networks (Hypothesis 4). In particular, social networks with maximum information
entropy (completely heterogenous ties) encourage innovation at almost three times the
rate of networks with no entropy (completely homogenous ties). Considering the other
structural covariates, it is worth noting that the inclusion of the diversity variable
slightly dampens the statistical significance of discourse-directed ties. Quite possibly,
one principal benefit of such directed ties—lack of information redundancy—is
addressed to some extent by the network diversity measure.
Since the data analyzed in Table 4 include both individual entrepreneurs and
entrepreneurial teams, only a few features of team composition can be addressed for the
full sample (see Model 4). As predicted, innovative propensity increases with the size of
entrepreneurial teams. When large numbers of entrepreneurs are brought together,
new combinations of ideas or routines become more likely (Hypothesis 5). Attempts at
innovation are also encouraged by a diverse set of functional roles within a team. The
multiplicity of role structures that entrepreneurs may invoke in a given context can
lead to role conflict (both inter- and intrapersonal), contributing to the modification of
conventional scripts (Hypothesis 6).
4.2 Applying for patents or trademarks
Results for a behavioral indicator of innovative propensity—patent or trademark
application—suggest a number of parallels with subjective perceptions of innovation
Strong ties, weak ties and islands 443
(see Table 5). Embeddedness in a particular industry tends to dampen attempts by
entrepreneurs to patent or trademark their ideas (Hypothesis 9), while such attempts
are encouraged by attention directed to public discourse as a source of ideas
(Hypothesis 3), as well as large entrepreneurial teams (Hypothesis 5). A number of
other estimates are consistent with predictions—including those for strong–weak ties
(Hypothesis 1), mimetic activity (Hypothesis 2), diverse advice networks (Hypothesis
4) and functional diversity (Hypothesis 6)—but these do not attain statistical
significance for this operationalization of the dependent variable.
One control variable that deviates from the analysis of subjective perceptions of
innovation involves serial entrepreneurs. Prior entrepreneurial experience increases the
propensity of startup founders to seek legal protection for their ideas in the form of
patents or trademarks, while it has no marked impact on the likelihood that founders
will view their ideas as being unusually creative (cf. Table 4). These empirical results are
consistent with the interpretation that serial entrepreneurs are relatively sophisticated
innovators, with greater awareness of other novel ideas in their business domain and,
moreover, awareness of how those ideas may pose competitive threats to their own
creative efforts.
4.3 Team composition
Table 6 considers the impact of team composition in greater detail. Because the
analyzed subsample is ‘biased’—comprising startup efforts involving at least two
entrepreneurs—only selected parameter estimates are shown. The impact of internal
team structure follows the rank ordering implied by hypothesized predictions (7 and 8),
Table 6 Selected estimates from ordered probit models predicting subjective measure of
innovation based on team composition
Variable Model 1
Cultural embeddedness
Years of industry experience –0.016 (0.007)**
Diversity of experience 0.019 (0.013)
Team composition
Number of entrepreneurs 0.029 (0.056)
Role diversity 0.043 (0.041)
Strong ties
a
–0.229 (0.179)
Weak ties –0.066 (0.200)
–2 log likelihood 1338.91
Number of cases 421
a
‘No ties’ represent the omitted structural category.
*P < 0.05; **P < 0.01; ***P < 0.001.
444 M. Ruef
with teams composed of strong tie networks being slightly less innovative than those
built on weak-tie networks and those composed of weak ties being less innovative than
those involving no prior relationships. However, differences in the rank ordering are
not statistically significant. In comparison to external ties, the modest effect of internal
ties probably results from the ‘reconstruction’ of tie strength once teams are formed.
While team members may have only limited familiarity with one another initially,
relatively frequent interaction during the formation of a new organization tends to
foster group integration. Consequently, pressures for conformity in the innovation
process may grow and largely obscure the initial benefit that interpersonal anonymity
holds for creative activity.
The parameter estimates indicate that the effects of team size and role diversity are
attenuated once attention is restricted to multimember entrepreneurial teams. Partially,
this may be attributed to data attrition and the resulting reduction in statistical power.
However, it is also possible that the principal benefit of team size—in terms of
encouraging creative action—occurs only as one moves beyond individual entre-
preneurs to multimember teams, with few marginal benefits accruing to the addition of
team members beyond that point. Similarly, the principal benefit of role diversity may
apply to individual entrepreneurs who assume several functional backgrounds, rather
than to heterogenous multimember teams. The analysis also suggests that industry
experience makes its main contribution through the average level of tenure among
founding team members, not by drawing on members with diverse industry experi-
ences from different cohorts.
5. Discussion
The results provide strong support for the thesis that capacity for business innovation is
both enabled and constrained by existing social structure. While treatments of
innovation and discovery often suggest that new combinations of ideas be treated sui
generis as largely random occurrences among isolated actors (e.g. Popper, 1959), these
undersocialized conceptions ignore the importance of embeddedness in triggering
combinations of ideas. Embeddedness is crucial to the flow of non-redundant
information to innovative entrepreneurs, particularly those who are able to draw on
weak ties, directed ties or non-equivalent role structures. At the same time, propensity
for innovation requires that entrepreneurs not fall prey to the conformity that might be
encouraged by social embeddedness. As a long-standing research tradition on inter-
personal expectations (e.g. Blanck, 1993) has emphasized, deference to the opinions of
others can serve as a significant constraint on creative experimentation. Our results
suggest that entrepreneurs can avoid the pitfalls of conformity by diversifying their
networks to include a wide range of social contacts and by emphasizing abstract con-
ceptions of ideas rather than concrete implementations.
In many respects, the cultural embeddedness of entrepreneurs appears to be as
important to their innovative tendencies as their structural embeddedness. Extensive
Strong ties, weak ties and islands 445
experience in an industry leads to predictability of means and achievement of ends,
inhibiting organizational innovation in the process. While ‘oversocialization’ is thus
inimical to innovation, the internalization of norms and ideas cannot be ignored from
an empirical standpoint in predicting capacity for creative action (cf. Granovetter,
1985). The internal structure of an emerging venture also factors into the tendency for
it to deviate from established organizational forms or to reproduce them. In contrast to
structural and cultural embeddedness, the direction of causality seems less clear on
this point. Do large entrepreneurial teams (or diverse functional roles) encourage
innovation? Or do the risks inherent in an innovative business plan push entrepreneurs
to expand their teams and incorporate additional functional backgrounds? In this case,
it seems more appropriate to talk about the ‘co-evolution’ of internal organizational
structure and innovation, rather than one-sided causal determination. More precise
timing data on the formation of entrepreneurial teams and codification of business
ideas can be examined to disentangle the relationship. Additional timing data is also
desirable to clarify the interplay of initial network ties within a team, innovation, and
the tendency toward team integration.
6. Conclusion
While the analyses presented here are a first step in addressing the embeddedness of
innovation, further empirical development is needed to connect these findings to
industrial evolution. In contrast to the micro-level view of innovation advanced in this
article, the study of industrial evolution requires an understanding of (i) the way that
innovation may actually translate into durable changes in market structure; and (ii)
whether and how such changes come to be recognized by actors outside the social
sphere of the innovator. Much that passes for ‘innovation’ from the standpoint of
an entrepreneur will seem little more than a flight of fancy (or madness) from the
standpoint of his or her contemporaries. Indeed, the majority of business innovations
seem doomed to fates of failure or cultural amnesia. The intersection of perspectives on
embeddedness and diffusion processes promise to contribute further insights to our
understanding of how innovations become durable features of the industrial landscape.
From a substantive viewpoint, it is also important to clarify how the micro-dynamics
of organizational innovation, in particular, relate to the macro-dynamics of the
emergence of novel organizational forms (see Hannan and Freeman, 1986; Ruef, 2000).
Researchers have suggested that, ‘among entrepreneurs and the new ventures they
create, we mostly find mundane replications of organizational forms’ (Aldrich and
Kenworthy 1999). Even when deviations from these ‘mundane replications’ do occur,
this does not automatically translate into the emergence of a new form. Many
innovative ventures disband before they capture any public attention that serves to
legitimate their distinctiveness. Others persist but fail to activate the social movements,
technological imitation and/or regulatory transformations needed to constitute the
identity of a new organizational form on a wider scale (Ruef, 2000). In these respects,
446 M. Ruef
the relationship between organizational innovation and the emergence of new organ-
izational forms serves as a special case of broader theoretical questions connecting
creative action and corporate transformation.
Acknowledgements
This research was supported by the Center for Entrepreneurial Studies at the Stanford
Graduate School of Business. The author would like to thank Howard Aldrich, Bill
Barnett, Valery Yakubovich and anonymous ICC reviewers for suggestions related to the
article, as well as seminar participants at INSEAD and the Stanford Graduate School of
Business.
Address for correspondence
Martin Ruef, Department of Sociology, Chapel Hill, NC 27599-3210, USA. Email:
ruef@email.unc.edu.
References
Ahuja, G. (2000), ‘Collaboration networks, structural holes, and innovation: a longitudinal
study,’ Administrative Science Quarterly, 45, 425–455.
Alchian, A. (1950), ‘Uncertainty, evolution, and economic theory,’ Journal of Political Economy,
58, 211–221.
Aldrich, H. (1999), Organizations Evolving. Sage: Thousand Oaks, CA.
Aldrich, H. and A. Kenworthy (1999), ‘The accidental entrepreneur: Campbellian antinomies and
organizational foundings,’ in J. Baum and B. McKelvey (eds), Variations in Organization
Science: In Honor of Donald T. Campbell. Sage: Newbury Park, CA.
Bian, Y. (1997), ‘Bringing strong ties back in: indirect ties, network bridges, and job searches in
China,’ American Sociological Review, 62, 366–385.
Blanck, P. (ed.) (1993), Interpersonal Expectations: Theory, Research, and Applications. Cambridge
University Press: New York.
Boorman, S. (1974), ‘Island models for takeover by a social trait facing a frequency-dependent
selection barrier in a Mendelian population,’ Proceedings of the National Academy of Sciences,
71, 2103–2107.
Burt, R. (1987), ‘Social contagion and innovation: cohesion versus structural equivalence,’
American Journal of Sociology, 92, 1287–1335.
Burt, R. (1992), Structural Holes: The Social Structure of Competition. Harvard University Press:
Cambridge, MA.
Carroll, G. and M. Hannan (2000), The Demography of Corporations and Industries. Princeton
University Press: Princeton, NJ.
Coleman, J. S., E. Katz and H. Menzel (1966), Medical Innovation. Bobbs-Merrill: Indianapolis.
Strong ties, weak ties and islands 447
Damanpour, F. (1988), ‘Innovation type, radicalness, and the adoption process,’ Communication
Research, 15, 545–567.
Damanpour, F. (1992), ‘Organizational size and innovation,’ Organization Studies, 13, 375–402.
DiMaggio, P. and W. W. Powell (1983), ‘The iron cage revisited: institutional isomorphism and
collective rationality in organizational fields,’ American Sociological Review, 48, 147–160.
Giddens, A. (1984), The Constitution of Society. University of California Press: Berkeley.
Granovetter, M. (1973), ‘The strength of weak ties,’ American Journal of Sociology, 78, 1360–1380.
Granovetter, M. (1985), ‘Economic action and social structure: the problem of embeddedness,’
American Journal of Sociology, 91, 481–510.
Granovetter, M. (1995), Getting a Job: A Study of Contacts and Careers, 2nd edn. University of
Chicago Press: Chicago, IL.
Greene, W. (1998), LIMDEP Version 7.0. Econometric Software, Inc.: Bellport, NY.
Hage, J. T. (1999), ‘Organizational innovation and organizational change,’ Annual Review of
Sociology, 25, 597–622.
Hannan, M. and J. Freeman (1984), ‘Structural inertia and organizational change,’ American
Sociological Review, 49, 149–164.
Hannan, M. and J. Freeman (1986), ‘Where do organizational forms come from?’ Sociological
Forum, 1, 50–72.
Hedström, P., R. Sandell and C. Stern (2000), ‘Mesolevel networks and the diffusion of social
movements: the case of the Swedish Social Democratic Party,’ American Journal of Sociology,
106, 145–172.
Ibarra, H. (1993), ‘Network centrality, power, and innovation involvement: determinants of
technical and administrative roles,’ Academy of Management Journal, 36, 471–501.
Malerba, F. and L. Orsenigo (1996), ‘Technological regimes and sectoral patterns of innovative
activities,’ Industrial and Corporate Change, 6, 83–117.
March, J. (1991), ‘Exploration and exploitation in organizational learning,’ Organization Science,
2, 71–87.
Marsden, P. (1987), ‘Core discussion networks of Americans,’ American Sociological Review, 52,
122–131.
Marsden, P. and K. Campbell (1984), ‘Measuring tie strength,’ Social Forces, 63, 482–501.
Miner, A., S. Ragahavan and P. Haunschild (1999), ‘Models of imitation and the convergence of
routines in a population,’ in J. Baum and B. McKelvey (eds), Variations in Organization Science:
In Honor of Donald T. Campbell. Sage: Newbury Park, CA.
Podolny, J. and T. Stuart (1995), ‘A role-based ecology of technological change,’ American Journal
of Sociology, 100, 1224–1260.
Popper, K. (1959), The Logic of Scientific Discovery [1934]. Hutchinson: London.
Powell, W., K. Koput and L. Smith-Doerr (1996), ‘Interorganizational collaboration and the locus
of innovation: networks of learning in biotechnology,’ Administrative Science Quarterly, 41,
116–145.
Renzulli, L., H. Aldrich and J. Moody (2000), ‘Family matters: gender, networks, and entre-
preneurial outcomes,’ Social Forces, 79, 523–546.
448 M. Ruef
Reynolds, P. and S. White (1997), The Entrepreneurial Process: Economic Growth, Men, Women,
and Minorities. Quorum Books: Westport, CT.
Rogers, E. M. (1995), Diffusion of Innovations, 4th edn. Free Press: New York.
Ruef, M. (2000), ‘The emergence of organizational forms: a community ecology approach,’
American Journal of Sociology, 106, 658–714.
Saloner, G., A. Shephard and J. Podolny (2001), Strategic Management. John Wiley: New York.
Schumpeter, J. (1934), The Theory of Economic Development [1911]. Harvard University Press:
Cambridge, MA.
Sewell, W. (1992), ‘A theory of structure: duality, agency, and transformation,’ American Journal
of Sociology, 98, 1–29.
Shannon, C. and W. Weaver (1963), The Mathematical Theory of Communication [1949].
University of Illinois Press: Urbana, IL.
Sørenson, J. and T. Stuart (2000), ‘Aging, obsolescence, and organizational innovation,’
Administrative Science Quarterly, 45, 81–112.
Stewart, A. (1989), Team Entrepreneurship. Sage: Newbury Park, CA.
Strang, D. and J. Meyer (1993), ‘Institutional conditions for diffusion,’ Theory and Society, 22,
487–511.
Strang, D, and S. Soule (1998), ‘Diffusion in organizations and social movements: from hybrid
corn to poison pills,’ Annual Review of Sociology, 24, 265–290.
Strang, D. and N. B. Tuma (1993), ‘Spatial and temporal heterogeneity in diffusion,’ American
Journal of Sociology, 99, 614–639.
Sundbo, J. (1998), The Theory of Innovation: Entrepreneurs, Technology and Strategy. Edward
Elgar: Northampton, MA.
Uzzi, B. (1996), ‘Embeddedness and economic performance: the network effect,’ American
Sociological Review, 61, 674–698.
Uzzi, B. (1999), ‘Social relations and networks in the making of financial capital,’ American
Sociological Review, 64, 481–505.
Weber, M. (1968), Economy and Society: An Interpretive Sociology [1924], ed. G. Roth and C.
Wittich. Bedminster: New York.
Zavoina, R. and W. McElvey (1975), ‘A statistical model for the analysis of ordinal level dependent
variables,’ Journal of Mathematical Sociology, 4, 103–120.
Strong ties, weak ties and islands 449
